Reducing Correlation for Confirmatory Sampling Location Identification During a Contamination Event

<p dir="ltr">The use of confirmatory sampling to supplement a fixed-sensor network may improve the characterization of a potential ongoing contamination event and improve response actions. In previous research, a cluster-based approach was proposed to place a set of confirmatory sampling locations (CSLs) in a reduced solution space to ensure efficient placement for realistic networks. However, the sets of CSLs selected were placed in close spatial proximity implying information overlap. The current research explores the trade-off between Information Gain (IG) and the D-Optimality criterion seeking to maximize the information collected from a contamination event, while minimizing the overlap between sampling sites. The multi-objective problem was solved using the Multi-Objective Updating Greedy Algorithm (MUGA) a greedy heuristic approach using an approximation to the IG metric to solve the placement problem in a reduced amount of time. Overall, the best performance occurred when the D-Optimality objective was given a small amount of weight keeping the solution out of a local minimum. In general, the MUGA approach was shown capable of creating Pareto front of solutions within a time frame appropriate for real-time application.</p><p dir="ltr">This paper was presented at the 21st Computing and Control in the Water Industry Conference (CCWI 2025) at the University of Sheffield (1st - 3rd September 2025).</p>